Bearing Fault Diagnosis Using Heterogeneous Image Fusion and MobileViT

Fault diagnosis in electromechanical systems has been widely applied in the industrial field to enhance the reliability of equipment operation and production efficiency. Traditional fault diagnosis methods primarily rely on statistical analysis of sensor data and rule-based models, which often depend on a significant amount of manual experience and professional knowledge. This paper proposes a fault diagnosis method based on heterogeneous image feature fusion and the MobileViT network. By incorporating three heterogeneous image transformations of time series data, i.e. Short-Time Fourier Transform (STFT), Local Binary Patterns (LBP) and Scatter Symmetrized Dot Pattern (SDP) utilizing the advanced MobileViT network from the field of computer vision for training and testing, the experimental results demonstrate that the proposed fault diagnosis method has achieved satisfactory diagnostic outcomes. Notably, when multichannel signals are input into the model, the predictive accuracy of the model can be further improved, with an accuracy rate as high as <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 8. 8 \%}$</tex>.